如何将unicode输出到emacs消息缓冲区？

Question

If I run the code

# -*- coding: utf-8 -*-
month = "März"
print month.decode("utf-8")

in the OS X terminal, I get the string März just fine.

Also, my emacs (24.5 on OS X 10.10) seems to handle unicode (or at least umlauts) just fine, since I can see the umlaut in my emacs window.

Yet when I run the code above directly from within emacs I get:

Traceback (most recent call last):
  File "unicode-umlaut.py", line 3, in <module>
    print month.decode("utf-8")
UnicodeEncodeError: 'ascii' codec can't encode character u'\xe4' in position 1: ordinal not in range(128)

What does this mean? Does it mean that even though emacs is handling a latin-1 character, the emacs Message buffer refuses to handle unicode? Is there a fix to make it possible to output non-ascii characters to the Message buffer in emacs?

Update:

Byte-wise the file looks (via emacs hexl-mode) like this:

00000000: 2320 2d2a 2d20 636f 6469 6e67 3a20 7574  # -*- coding: ut
00000010: 662d 3820 2d2a 2d0a 6d6f 6e74 6820 3d20  f-8 -*-.month = 
00000020: 224d c3a4 727a 220a 7072 696e 7420 6d6f  "M..rz".print mo
00000030: 6e74 682e 6465 636f 6465 2822 7574 662d  nth.decode("utf-
00000040: 3822 290a                                8").

The c3a4 maps to a-umlaut (ä), and so the file seems to be properly coded in UTF-8.

Answer 1

This:

# -*- coding: utf-8 -*-
month = "März"
print month.decode("utf-8")

is more simply:

# -*- coding: utf-8 -*-
month = u"März"   # Use a Unicode string!
print month

#coding: utf8 declares the encoding of the source file , so make sure your editor is configured to save the file in that format.

The first way would break if run on a terminal not configured for UTF-8; the second will work on a terminal configured for any encoding that supports the ä character.

The error message you've shown indicates month is already Unicode, so Python 2 is trying to encode it with the default ascii codec before decoding it back to Unicode with the utf8 codec. That implies you are not running the same code displayed above, since that code uses a byte string.